bmad-os-findings-triage

Automated triage of code review findings uses parallel AI agents to research and propose resolutions for each finding while keeping humans in the loop for final decisions. This orchestrates the analysis to clear review backlogs efficiently.

Human-in-the-loop triage works by having agents present analyzed findings to humans who then issue binding decisions like FIX, DISMISS, or DEFER. The system records these decisions and updates GitHub PR comments automatically.

Yes, integrating with GitHub CLI allows the triage system to post results and update PR comments automatically. It manages the lifecycle of findings from task creation to final decision recording.

Yes, local task management tools are required alongside GitHub CLI to execute autonomous research and commit verified fixes. These integrations manage the lifecycle of findings from parsing to final commit.

The best way to handle parallel agent orchestration is spawning individual agents for each code review finding. This ensures focused, high-quality analysis before presenting results for human decisions and automated GitHub reporting.

Automated triage requires a findings report file to parse and create initial tasks for the agents. The system reads the report, spawns agents to analyze each finding, and manages the workflow until fixes are committed.